EXECUTIVE SUMMARY
Style & Statistics: The Art of Retail Analytics
Five Hot Trends in
Retail Analytics
Smart retailers are
using analytics by
harnessing the big
data streams that
define the digital
nature of our increasingly online, mobile
world to inform
more meaningful
omnichannel
strategies.
Retail isn’t an easy place to be these days.
The environment is omnichannel and everchanging. Competition is rising, and retailers
are struggling to understand how to best
meet customers’ merchandise preferences
and drive profitability and market share.
Smart retailers are doing that with analytics
by harnessing the big data streams that
define the digital nature of our increasingly
online, mobile world to inform more meaningful omnichannel strategies. Let’s take
a look at the five hottest trends in retail
analytics.
Trend 1: Volume and
climate clusters are so
#oldschool
Remember when the only option for
shopping was to get into your car and drive
to the store? Those days are long gone.
Shoppers often engage with brands online
before even stepping into the store, and
many continue on to purchase online.
Personally, I typically window-shop from my
couch in comfy yoga pants.
To understand customers’ merchandise
preferences, it’s critical to break down the
silos of channels. Look across channels to
understand what products individuals are
purchasing online, what products they’re
purchasing in store, and what products have
an opportunity within a local market.
Retailers need to use analytically driven clustering techniques to cluster local markets or
trade areas by similar merchandise preferences and selling patterns. It’s one smart
way to use retail analytics to stay relevant
with our omnichannel customers.
Trend 2: Attribute analytics
are all the rage
Merchandise attributes are ways to describe
products such as color, sleeve length, silhouette, fabric and so forth. Retailers can analyze
any number of merchandise attributes and
vary this analysis by area.
More attributes are not always better.
Analyzing too many attributes can become
overwhelming, and quite frankly, the juice
may not be worth the squeeze. Correlation
analysis can be used to understand which
merchandise attributes are correlated. This
can help narrow the number of attributes
down to a manageable size.
But what draws customers into the department and what drives their purchasing decisions may also be understood through
attribute analytics.
Retailers need to use analytically driven
clustering techniques to cluster local markets
or trade areas by similar merchandise preferences and selling patterns
I always head toward the Jessica Simpson
heel display because I know the price is
right, the styles are cute and the shoes are
actually comfortable. I can’t wear heels that
aren’t comfortable, no matter how cute they
are. I end up looking like a baby deer trying
to walk in stilettos. It’s not a good look. Now,
I know that brand is driving my purchasing
decisions for heels, but what about every
other customer?
You can use analytics to statistically determine what merchandise attributes are the
most important to customers – and then,
within those attributes, the specific values of
those attributes that have opportunity.
We can understand, for example, whether it
is brand or fit that drives customers’ decisions and then, within that, which brands or
fits have the most opportunity by cluster.
Retailers can use these insights to drive
assortment planning decisions and move
from product-centric to customer-centric
assortments.
By operationalizing their analytics, the savvy
retailer uses the data that represents the
attributes, discovers the most relevant attributes, and then deploys those insights
through assortment and allocation decisions
to deliver the best experience for their
customers.
Trend 3: Manual sales
planning is out, statistical
forecasting and price
optimization are in
Retailers can throw out their sticky notes
reminding them of previous events. Clear
their calendar reminders for past promotions.
Let merchants focus on the art of analytics.
Different events, causal factors, promotions,
merchandise and stores can create a big
data problem. It’s about using analytics to
drive decisions from big data with greater
accuracy. Using a statistical forecast enables
retailers to predict future demand. For new
The statistical forecasting process automates the
ability to account for promotions and holidays,
giving merchants back more time in their day.
items and previous items, analytics can
determine the significance of events and
promotions and account for outliers and
anomalies in the data. The statistical forecasting process automates the ability to
account for promotions and holidays, giving
The other aspect here is the use of imputation techniques. Imputation might sound
like a word out of a high school biology or
genetics textbook, but don’t be scared. In
statistics, it’s the process of replacing missing
data with values. In this example, imputation
merchants back more time in their day.
is replacing missing sales due to a lack of
inventory. Each time a store sells out of a
size, it creates what is referred to as an out of
stock. When stores have items out of stock,
they miss out on sales. Sometimes this
concept is referred to as lost sales.
Statistical methods have also been proven
to increase forecast accuracy by more than
50 percent. The increase can equate to a
significant optimization in inventory, which
leads to an increase in gross margin and
profitability.
Gone also are the days of using business
rules and product age to drive promotion
and markdown decisions. Leading retailers
are taking advantage of analytics to understand customers’ price sensitivity and to
determine the optimal recommendation
down to a customer or store level. Localized
pricing strategies can drive 5 percent to 15
percent of incremental margin – profit that
retailers are otherwise leaving on the table.
Trend 4: Steer clear of the
loser loop
Retailers can have the most amazing assortment, completely tailored to customer preferences. But if the retailer doesn’t have the
customer’s size in stock, it’s a lost sale – an
epic fail. Traditionally, historical point-of-sale
data has been used to determine what the
size curve of a given product should be for
locations. The problem is that if they didn’t
stock the size, then they didn’t sell the size,
ultimately creating a loser loop.
However, more advanced statistical imputation techniques can create a much more
accurate result. For example, being able to
take into account when in the product life
cycle a stock-out occurred gives a much
more intelligent perspective. If the location
didn’t have inventory on hand, the next step
is determining where the item was in its
product life cycle. If the item was out of stock
at a meaningful point in its life cycle, the lack
of inventory could have affected the
demand. The earlier in the life cycle this
occurs, the greater the effect. But if the item
was about to be marked down to clearance
the next day, the fact that it was out of stock
is actually a good thing.
Using analytics to determine the true size of
demand for a given product down to a
location level will ensure that customers find
the right product in the right size. This
process not only creates a better customer
experience but also avoids markdowns,
increasing overall profitability.
Trend 5: Personalization
The relationship between retailers and
customers should be viewed as a two-way
street. Customers are continuously giving
retailers information in real time through
their browsing and buying behavior – online
and off-line. Today’s breed of customer
expects that retailers will use this information
to offer them the most relevant
communications.
For example, women’s apparel retailer
Chico’s FAS gathers information at many
points – from customer-provided information to website activity to social media and
in-store visits and purchases – to create
a continuously evolving model of each
customer. That’s each individual customer.
The customer can direct where and when
she decides to receive contacts, when and
how to act. She can do this anywhere,
anytime, on any device, and the experience
will be consistent. Information gained from
each channel is reflected in every other
channel – never lost or fragmented.
The retailer’s Digital Retail Theatre concept
and model makes omnichannel retailing
a reality and takes it to a diversity of digital
devices such as interactive displays, mobile
phones and tablets. Far more than just a
modified website experience, it is retailing
redefined. Boundaries are erased. Lines are
blurred. There is no start or end to the experience. Customers have unprecedented
options in their shopping experiences as
well as their relationship with the retailer –
delivering an omnichannel experience
that is powerful for its real-time relevance.
Where technology was once viewed as
depersonalizing the human experience,
analytics is bringing personalization back
in a new way.
Learn more
Learn more about how retailers are using
analytics to redefine how they interact with
customers, provide the right communications to suit their preferences, and succeed
in the new omnichannel world in my book,
Style & Statistics: The Art of Retail Analytics.
For insights from MIT Sloan Management
Review research on how the retail industry
is using analytics, start with the industry brief,
The Analytics Mandate: Helping Retailers
Evolve.
About the author
Brittany Bullard is a systems engineer for the retail and
consumer packaged goods industries. She has nine years of
experience in the retail industry – implementing, managing
and designing solutions for allocation, replenishment, forecasting, location planning and assortment planning. At SAS,
Bullard helps retailers drive profitability through the use of
powerful analytics.
To contact your local SAS office, please visit: sas.com/offices
SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries.
® indicates USA registration. Other brand and product names are trademarks of their respective companies. Copyright © 2017, SAS Institute Inc.
All rights reserved. 108770_G43818.0317